Quote Originally Posted by HDNitehawk View Post
Anyone want to take a shot at explaining this:
Are we splitting pixels 4 ways and why would that even matter.
Wouldn't you just end up with 4 pixels together with the exact same makeup?
In-Camera Upscaling

For instances when 24MP isn't enough, DIGIC Accelerator processing now enables In-Camera Upscaling, which uses deep learning to double the number of pixels both horizontally and vertically in order to produce a separate 96MP image in about 10 seconds. This upscaling can be done after the original image is shot and can also be performed at the time of cropping images in camera, saving time compared to generating a 96MP image and then cropping it afterward.
You pulled from the R1 description, but in a nutshell this is just fancy upscaling. Photoshop does it (old school, with bicubic smoother interpolation), Topaz Gigapixel AI does it (with AI), and now Canon does it (with deep learning, which is probably similar to what Topaz does but more accurately named by Canon), too. Yay.